Dynamic computed tomography (CT) is widely used in humans to determine tumor perfusion via time-attenuation curves. Five types of time-attenuation curves have been identified and shown to have prognostic relevance in humans. The goal of our study was to assess the feasibility of this technology in spontaneous canine tumors and to determine time-attenuation curves and perfusion patterns in different tumor types. Thirty-one dogs with tumors accessible for biopsy were evaluated (15 carcinomas, 16 sarcomas). Dynamic CT was performed at the level of the largest tumor cross-section. Time-attenuation curves were calculated and ratios from the tumor to a contralateral artery were derived for wash-in, peak attenuation, time to peak attenuation, wash-out, and perfusion. Median perfusion was significantly higher and median time to peak ratio was significantly shorter in carcinomas and bone sarcomas compared with soft tissue sarcomas (P = 0.03 and 0.01). There was a trend of lower median upslope and wash-out ratio in soft tissue sarcomas in comparison with carcinomas (P = 0.06 and 0.09). Although peak ratio was lowest in soft tissue sarcomas, differences were not significant (P = 0.3). The most common type of time-attenuation curve for all tumors had a slow to moderately rapid wash-in with a low to moderate attenuation peak followed by a plateau phase. In conclusion, dynamic CT is feasible and time-activity curve-derived measurements differed between spontaneous canine tumors. More data has to be collected in a larger number of patients and correlated with response to treatment and outcome.